How do we Perform Failure Analysis using the CMMS?

In the above figure, the Asset Manager is asking some tough questions.?He just finished talking with corporate [they chewed him out for increasing O&M costs] and was not able to explain why breakdowns were increasing and impacting production.?He did run some graphical failure reports (from Excel) but these generated more questions than answers. It seems that meaningful failure analytics and analysis was not possible within the CMMS as the failure data - and report design did not exist.

Are we Asking the Right Questions?

Is there a more significant process than failure analysis? [Answer is no] And do we even have a CMMS utilization plan, which if existed, would state that we should use the CMMS to perform failure analysis? And by the latter, is it understood by all that this does not mean printing out individual work orders to find useful information?

How Much Failure Data is too Much versus too Little?

What is the right amount of failure data on a given work order? If you go too far, this exercise becomes a root cause failure analysis (RCFA) and becomes quite exhausting to analyze. But if you do too little, then it does no one any good, i.e. only capturing asset problem code. The author believes the ideal process is chronic failure analysis, Pareto-style, where we focus on recurring failures with the greatest impact to performance, risk to production, and cost to maintain. The following graphic shows the detail needed on a work order (with functional failure).

Using RCM Vernacular inside the CMMS

RCM analysis captures several data elements. And one of those elements is the failure mode. If this phrase can be segmented into 3 pieces, then it can be used for validation purposes. Further, if the CMMS work order can also capture a validated failure mode (in 3 pieces), then we have a way to perform electronic comparison.

Sorting the Bad Actors

When running a bad actor report, you need to first select those work orders with functional failures. The next step is to sort the list. Using validated data, a programming language can aggregate data (i.e. group data), sort on highest group counts, and automatically stop at cut-off point, e.g. 10. This process quickly identifies the bad actors. The sort metric might be Average Annual Maintenance Cost divided by Replacement Cost. And once an asset is selected as bad actor, then the review team drills down on the failure mode to arrive at cause. Imagine the power!

What Kind of Meeting Would You Prefer?

Does this Data Capture Require More Time by Techs?

Answer: probably yes, but not much. If the techs were in the conversation, they might complain about the time to enter failure data. I might counter by saying, our goal here is to reduce reactive maintenance. The cost benefit is huge when compared to the additional time to enter validated failure data.

Pursuing Operational Excellence

As Joseph Paris defined operational excellence, "All efforts in the organization are in a state of alignment".

The book, Failure Modes to Failure Codes, defines this design in detail.